Hot off the heels of OpenAI releasing their GenAI chatbot ChatGPT to the public in November 2022, Google released their own chatbot called Bard (now Gemini). During Bard’s first public demonstration, it generated a major factual error in response to a question about the discoveries made by the James Webb Space Telescope. This wrong answer by the chatbot led to a 9% drop in the stock price of Alphabet, Google’s parent company — at the time, $100 billion in market value.
The Risks of Botshit
Don’t let inaccurate chatbot content harm your business.
July 17, 2024, Updated July 25, 2024
Illustration by HBR Staff
Summary.
Botshit — made-up, inaccurate, and untruthful chatbot content that humans uncritically use for tasks — can pose major risks to your business in the form of reputational damage, incorrect decisions, legal liability, economic losses, and even human safety. Yet, it’s unlikely that chatbots are going away. How can you manage these risks while taking advantage of the benefits of promising new tools? The authors suggest asking two key questions based on their research: How important is chatbot response veracity for a task? And how difficult is it to verify the veracity of the chatbot response? Based on your responses to these questions, you can better identify the risks associated with a given task — and successfully mitigate them.